Image decoding apparatus, image encoding apparatus, and image decoding method

Abstract

According to an aspect of the present disclosure, an image decoding apparatus for decoding information specifying a neural network includes: header decoding circuitry that decodes an input tensor identification parameter specifying a process for deriving an input tensor input for a post filter of the neural network. The input tensor identification parameter is a parameter related to a color component channel.

What is claimed is: 
     
       1. An image decoding apparatus for decoding information specifying a neural network, the image decoding apparatus comprising:
 header decoding circuitry that decodes an input tensor identification parameter specifying a process for deriving an input tensor for a post filter of the neural network, 
 wherein 
 the input tensor identification parameter is a parameter related to a color component channel. 
 
     
     
       2. The image decoding apparatus of  claim 1 , wherein a relational expression for deriving the input tensor is defined by the input tensor identification parameter and a chroma sampling of an input image. 
     
     
       3. The image decoding apparatus of  claim 1 , wherein the header decoding circuitry decodes an output tensor identification parameter specifying a number of color component channels of an output tensor of the post filter of the neural network. 
     
     
       4. The image decoding apparatus of  claim 3 , wherein a relational expression for deriving an output image by using the output tensor is defined by the output tensor identification parameter and a chroma sampling of the output image. 
     
     
       5. The image decoding apparatus of  claim 1 , wherein the header decoding circuitry decodes neural network model complexity information. 
     
     
       6. The image decoding apparatus of  claim 5 , wherein the neural network model complexity information includes a first syntax element indicating a number of neural network parameters for the post filter. 
     
     
       7. The image decoding apparatus of  claim 5 , wherein the neural network model complexity information includes a second syntax element indicating a parameter type of the neural network. 
     
     
       8. The image decoding apparatus of  claim 5 , wherein the neural network model complexity information includes a third syntax element indicating a bit size of the neural network. 
     
     
       9. The image decoding apparatus of  claim 6 , wherein the header decoding circuitry derives a maximum number of parameters by using the first syntax element. 
     
     
       10. The image decoding apparatus of  claim 5 , wherein the neural network model complexity information includes a fourth syntax element specifying a number of operations for the post filter. 
     
     
       11. An image encoding apparatus for encoding information specifying a neural network, the image encoding apparatus comprising:
 header encoding circuitry that encodes an input tensor identification parameter specifying a process for deriving an input tensor for a post filter of the neural network, 
 wherein 
 the input tensor identification parameter is a parameter related to a color component channel. 
 
     
     
       12. An image decoding method for decoding information specifying a neural network, the image decoding method including:
 decoding an input tensor identification parameter specifying a process for deriving an input tensor for a post filter of the neural network, 
 wherein 
 the input tensor identification parameter is a parameter related to a color component channel.